Four-dimensional magnetic resonance imaging-based system for radiotherapy of lung cancer
Date of Issue2014
School of Chemical and Biomedical Engineering
Lung cancer has been the most common and the most deadly cancer in the world for several decades. For current 3D-radiotherapy (RT) especially for lung cancer patients, respiration motion poses a major challenge. Consequently, four-dimensional (3D + time) RT, which involves the use of 4D imaging modalities, is developed to address this problem. The overall objective of this project is to significantly improve the efficiency and efficacy of RT. Currently, the 4D-RT planning of lung cancer is only based on 4D computed tomography (CT). However, 4D-CT involves high ionizing radiation and is only able to provide an averaged breathing cycle to study respiratory motion. As compared to 4D CT, magnetic resonance imaging (MRI) is able to produce images with excellent soft tissue contrast. In addition, there is no ionizing radiation involved during MRI scanning. Consequently, 4D-MRI is able to continuously scan in ‘real-time’ for several breathing cycles, which makes 4D-MRI able to cover individual changes in breathing. This thesis proposes the development of a novel 4D-MRI based system for 4D-RT of lung cancer patients as a complement of current 4D-CT based system. The specific aims of this thesis are (i) determining suitable 4D-MRI sequences for lung cancer imaging; (ii) developing automatic target structure delineation; and (iii) the fusion of 4D-MRI and 3D-CT for dose calculation. Ten healthy subjects and six lung cancer patients were recruited and studied in this thesis. 4D-MRI sequence TWIST was found to be suitable for dynamic lung imaging as it showed good image quality at a faster temporal resolution, which is capable of showing the motion path of tumor. A novel automatic registration-based segmentation scheme was successfully developed which was shown to greatly reduce computation amount/time while maintaining good segmentation accuracy. To address the problem of ionizing radiation and dose calculation, a novel technique was developed to fuse 4D-MRI and 3D-CT to generate simulated 4D-CT datasets. The simulated 4D-CT images were shown to be accurate in terms of landmark positions. In summary, the evaluation of 4D-MRI sequence, and the developed methods for target delineation of 4D datasets and fusion of temporal and spatial information from different modalities, provide a basis for the use of 4D-MRI as an alternative imaging modality to 4D-CT for 4D-RT of lung cancer patients.
DRNTU::Engineering::Computer science and engineering::Computing methodologies::Image processing and computer vision